(201e) Mitigation of Severe Slugging with Internal Model Control

In the offshore oil production field, unstable fluid flow may occur when two-phase fluid passes through a vertical riser connected to a downwardly inclined subsea line, or when a pipe is laid in a hilly terrain. This is called severe slugging or terrain induced slugging. As it passes through this section, liquid accumulates at the bottom of the pipeline and begins to block the passage through which the gas phase can pass, compressing the gas until it is pressurized to move the liquid. When the gas is able to overcome the gravitational head of the liquid, the accumulated liquid will come out at once. Mitigation of severe slugging is essential when designing an offshore plant because the liquid level of the separator may suddenly increase or vortex induced vibration (VIV) may appear in the riser due to the severe slugging.

The most common way to eliminate severe slugging is choking the topside valve. If the valve is choked under certain valve opening at the topside separator, back pressure and the velocity at the choke increases. This makes the liquid holdup to increase in the riser because more gas phase stream passes through valve than liquid phase stream. Positive perturbation of liquid holdup increases weight in riser and causes the liquid to fall down so that the pressure drop over the riser grows. Increased pressure drop increases the gas flow, allowing the gas to flush the liquid well and eliminating severe slugging.

Therefore, in this study, a multiphase flow loop which is consisted of downwardly inclined pipe and vertical riser in lab scale was designed and installed in order to generate and mitigate severe slugging. This flow loop consists of a vertical line of about 5.7m, a horizontal line of 12m, and a choke valve is installed on the upper part of the vertical line to mitigate the severe slugging. To compare the experimental results, model was made by inputting the dimensions of the flow loop using OLGA, a multiphase flow simulation program.

First, the opening of the topside valve was manually controlled to confirm that severe slugging was mitigated. A bifurcation diagram was made to more clearly identify which flow this system has at each valve opening. Bifurcation diagrams have been used to plot the values of pressure versus the values of valve opening for the slugging system. In the stable regions the plot consists of a unit curve showing the exact value of the pressure in simulations or the average of very small pressure oscillations in experiments. In unstable regions the plot consists of three curves, one for steady state conditions and the two others showing the maximum and minimum of oscillations at each value of valve opening over the work range of choke valve. Valve opening at the point where the severe slugging flow changes to stable flow is expressed as critical valve opening. Experimental results show that severe slugging disappears at the opening below the critical valve opening, and that various critical valve opening values ranging from 0.3 to 0.7 are obtained depending on the liquid and gas flow rates.

The OLGA model was tuned by adjusting the valve coefficient or the air tank diameter so that a bifurcation diagram similar to that of the experiment was obtained. Then, the model-based control logic was constructed by estimating the transfer function using the OLGA model. When estimating the transfer function, the inlet gas buffer pressure data near the critical valve opening were used and the System Identification Application provided by MATLAB was used. This application estimates the transfer function by numerical method based on the data calculated using OLGA. The transfer function is estimated to have two poles and one zero, and data from the OLGA data and the estimated transfer function show a high agreement rate of 80-90%.

The estimated transfer function is used to estimate the gain of the PI controller, using the method called the Internal Model Control (IMC). The IMC method is one of the model predictive control methods, which minimizes the difference between the transfer function of the actual system and the transfer function of the predicted model. The control gain is calculated by using this method and the calculated gain is input to the PI controller of the experimental equipment.

As a result, it was observed that stable flow can be obtained even if the valve opening is increased up to 0.45 from 0.33, which is the critical valve opening of Usg=0.7m/s, Usl=0.1 m/s condition when manual choking is performed. Likewise, it was also observed that stable flow can be obtained even if the valve opening is increased up to 0.9 from 0.65, which is the critical valve opening of Usg = 0.8m/s, Usl=0.2 m/s condition when manual choking is performed. When the PI controller gain is obtained through the transfer function of the flow loop system as described above, mitigation is possible even at the valve opening which is larger than manual choking. Therefore, it can be said that it is a very efficient control method because it can provide more production to topside while controlling severe slugging than manual choking.